Scale v24.04.2 Lost One disk, Unable to import Pool from 2nd one on new H/W

TrueNAS General
,
1

Hi,
The system was setup by a friend a month ago on the F2-423 Teramaster H/W with 32 GB RAM using TruNAS core. It was upgraded last week to Scale v24.04.2. This setup was running Plex for the home only and was working just fine. The TrueNAS config comprised of
1 pool

  • 1 Data VDEV - 2 discs (8 TB each, Non Raid/Stripe)
  • 1 Cache VDEV - 1 TB Nvme
  • 1 Boot VDEV - 128 GB Nvme

One of the two disc failed y’day. Storage was at 53% and the disc that failed had less than 4 % of the data. Most of the media is on the good disc

I moved everything to a different F2-423 (borrowed h/w), created a new pool and when attempting to import the existing pool from the one working drive, I don’t see anything in the drop down in the GUI under import pool.

When I plug in the good disc, I can see both pools, the new one and the previous one. The previous one however shows the Data VDEV as offline VDEVs

Is there any way I can import the pool from the existing good disc into the new pool and salvage that data or am I done.

Tried using zpool import -fF pool1 and got this…
cannot import ‘DATA’: I/O error
Destroy and re-create the pool from
a backup source.

This is my 1st foray into NAS so if I haven’t used the right vernacular or described something right, Pl. be patient and ask. I’ll be happy to provide more details

Thanks

SShot1
SShot11661×764 144 KB

SShot2

2

Is your two disk pool a stripe and not a mirror? If you lost a disk in that configuration, you data is gone. I hope you had a backup or can recreate the data.

How big was your two disk pool? About 8TB or 16TB?

3

Your friend made a mistake. If you use stripes and not RAID, all of your data is gone the moment any drive dies. It’s not salvageable. Only use non-RAID stripes, when you accept this, otherwise use RAID.

4

It sounds like they added the second, 8TB disk for more space.

Cache means L2ARC? You shouldn’t add that unless necessary (going by ARC hit ratio). You only have 32GB of RAM and a 1TB L2ARC is eating into your RAM usage for regular ARC. See ZFS Read Cache section linked below.

1 Like
5

If a stripe, then your only hope is:

Or recover the failed disk.

6

That is correct, No Raid, only Stripe. Last I checked, it was at 52%, just above 8 TB.

7

Understood. he has some experience running Raid systems at work and recommended TrueNAS with Stripe the H/W could only support two disks and and total data was around 10TB. What I am bummed about is that with one of the discs gone kaput, I lost data on the other too even though it was a non-raid config.

8

I am new to NAS so had no input in how it was config. My only ask was to stand up an OS capable of running Plex for our home media, that’s all.
I am learning now

9

Checked it out and only the paid version allows possible recovery. I cant justify the license cost of $500 since I have backups of my data/media which are just scattered amongst multiple storage devices, online & offline. It will be a 7-10 days worth of effort to curate it all again and get it up and running. Just need to find the time to do it.

Now that I understand the implications of using NAS a bit better, this may not be suited for my use-case. All I need is a light wt OS for a Plex server so will explore another os, a linux disto or even a lighter version of Windows 10/11. It is for home use only, configured on a vlan, with no external access so security is not a real concern.
Once I have everything stood up, I will clone the hosted disc as a one time backup in-case the original one goes kaput.

10

well even if you used different os, if you don’t plan your storage with some sort of redundancy you could end up with the exact same outcome on a different OS… why not just redo your truenas with a redundand setup and be done with it?

1 Like
11

Sorry - you pretty much need to rebuild you data pool.

  1. As previously pointed out if you lose a data vDev in a pool, that pool is pretty much gone. So whilst a mirror (or other redundancy) divide the chance of losing any data (you need to lose two - or more - drives before you lose any data), equally non-redundant stripes actually multiply the risk of data loss (you lose all your data the moment one drive has a problem).

  2. The cost of redundancy is greater with a mirror (50%) than with (say) a 4x RAIDZ1 (25%). A two slot NAS box (like the F2-423) should really only be used as a 2x mirror. (I have an F5-221 so I understand Terramaster hardware - albeit older.)

So - how do you fix this?

  • Accept that the disks you have need to be a 2x8TB mirror giving you 8TB of usable space; or
  • Buy 2x 16TB disks for a mirror giving you the previous 16TB of usable space and sell or reuse the 8TB drives.

You might also want to experiment as to whether you really need to store 4K video, and whether you actually notice the difference if you store 1080p instead (you will save a massive amount of disk space if you decrease the resolution).

In addition, IMO you should repurpose the 1TB NVMe (or swap it for a smaller one) as an SSD pool (backed up to HDD using ZFS dataset replication) so that you can host both the Plex app and the Plex metadata on SSD rather than on HDD. To assist with sizing, my own Plex Metadata is c. 60GB on c. 4TB of multimedia mostly 720p.

1 Like
12

  1. If you are thinking of regular disk clones, then you might as well stick with TrueNAS - a mirror and snapshots is a far better solution than disk cloning.

  2. If you are happy with non-redundant data, then just configure the 2x8TB drives as two separate pools so that you don’t lose all the data if one drive fails. The downside to this is that you will need to manage the split of data across the two pools yourself e.g. put TV series on one and films on the other.

    You can definitely do this with a manually built Linux or Windows install, but equally TrueNAS can achieve this without the need to change.

1 Like
13

You make good points. I can certainly limit what goes on Plex bringing it down to less than 7.8GB but also like the idea of creating separate pools for each disc thus isolating the loss. I do plan to bump up the disc size to 20GB each but that wont happen until next year.

Let me sit on this until this weekend and think it through. Appreciate you all taking the time to share your input.

Thanks

14

One of the biggest benefits of TrueNAS and ZFS is the ability to protect your data from drive failure and bitrot.

Drives fail. The data on a drive also randomly corrupts (bitrot)

Having redundancy means you can withstand a drive failure without any data loss, and zfs can automatically correct bitrot (when read, or as part of a scrub)

15

Another benefit of ZFS is that upon a bad block, (or multiple bad blocks), ZFS will tell you what files are impacted. What this means is that you only need to restore that file or files to bring the pool back to full function.

Naturally if you loose an entire disk in a non-redundant pool, the pool is like bread overly exposed to heat, (aka toast :smile: ).

One other feature of ZFS. By default, ZFS stores 2 copies of metadata, (like directory entries), and 3 copies of critical metadata. This means that if one copy is found to be bad, (disk block loss), ZFS will automatically use another copy. AND then fix the bad copy, (assuming your storage device still has spare blocks available).

This means a single bad block generally won’t cause loss of an entire file, (unless it is a tiny file and that block was the entire file). ZFS considers directory information, (and other more critical metadata), more important than regular data. Thus, the extra copies, which won’t really take that much extra space.

Proper use of ZFS makes it an incredibly robust file system, volume manage and RAID scheme all in one software package.

Linux may use 3 separate packages for the same function, (EXT4, LVM & MD-RAID), and still not have as much functionality as OpenZFS.

I use a non-redundant striped pool for my miniature media server. Over the 10 years I've had it, perhaps 2 dozens files have been impacted. All fully available in my backups and easily restored to bring my media pool back to 100% functionality.

Of course, if or when I replace my media server, I will attempt to use redundant media, (likely a simply 2 way Mirror).

1 Like
16

Sorry you friend didn’t set you up correctly or didn’t explain that the loss of one drive meant the loss of all data.

If you are building a media library, ZFS may be too costly for you.

Best of luck to you with whatever you do next.

17

Just make sure to remember to keep utilization under 80% of total pool capacity. If you got a 10TB drive, keep total used space under 8TB. Getting past 80% performance starts to drop, 90% performance drops HARD, 100% utilization; pool can completely freeze & even deleting a file to make space will be a PITA.

18

An 8TB drive will only allow you to store c. 6TB of data - max ZFS occupancy is 80% and other overheads will eat into the total space.

19

If the use is static media, basically just adding files, then you can go beyond 80% full.

One issue over 80% is fragmentation of the free space. (And of the files themselves…) But, if you are just adding files until close to full, 95%, then their is less of an issue.

Note what I mean by adding files, is that you don’t have file churn, deleting and adding. If that is the case, then 80% maximum full is a better choice.

Or you can start with a empty single disk pool and only add media files to it.

20

The plan is to continue with TrueNAS with the following topology (for now)

  1. Three Pools, I each for two 8TB discs and one for 1 256GB Nvme (downgraded from 1TB)
  2. Split the media across two 8GB discs with around 60-70% fill rate
  3. Use Nvme for Plex App

We are sifting through our media and segregating what on each of the two discs. Ballpark # for all essential media is around 7.5TB. We will end up with 5.5-6TB on one and remaining on the other.
This data is more or less static (photos, movies etc.) so disc churn is not concern.

Around XMas, we will upgrade the discs capacity to 20 or 22TB each and switch the topology to RAID1. Upgraded disc capacity should give all the space we need for the next few years.

Thanks again all for your guidance, recommendations & insightful feedback.